Revision as of 20:01, 9 April 2016

Project 2: 3D Models

From this point on we're no longer using the rasterizer code. In all future homework projects we're going to use OpenGL for all our rendering.

In this homework assignment you're going to learn how to:

• load polygonal OBJ files
• render polygons in modern OpenGL
• automatically center and scale 3D models
• set up the virtual camera
• control the camera with the mouse
• set up light sources
• set material properties

1. 3D Model Loader

You already know how to load point clouds. It turns out that the 3D model files from project 1 (Bunny, Bear, Dragon) contain surface descriptions as well, by means of triangle connectivity.

f 31514//31514 31465//31465 31464//31464

Modify your OBJ loader so that it also parses the face lines, then modify your code to display triangles instead of vertices for the OBJ objects. Afterwards, in order to make mouse controls you are about to implement work well, you will need to center your OBJ models and standardize their sizes. Write function calls to do both of these things whenever a new model is selected with the function keys. All models should get scaled to the same size, and should fully occupy the graphics window, without exceeding it.

2. Rendering using modern OpenGL

The starter code has been modified so that the Cube now gets rendered with modern OpenGL, through the use of a VAO (Vertex Array Object), VBOs (Vertex Buffer Object), and EBOs (element buffer object). In order to make the use of the programmable pipeline possible, a shader class was added to the starter code, along with a sample vertex and fragment shader. At the moment, the sample fragment shader causes all objects to be colored light orange. You will be fixing this later when you begin working with lights. Using Cube's example of the modern OpenGL approach, add the ability to render your OpenGL using modern OpenGL. Since you are no longer using the Rasterizer, use the 't' key to toggle between immediate mode (the OpenGL mode from project 1) and modern OpenGL. There should be a noticeable application speedup when using modern OpenGL.

3. Mouse control

It is time to support the mouse to control your 3D models. Add functionality to allow the following operations on the displayed 3D model:

• While the left mouse button is pressed and the mouse is moved, rotate the model about the center of your graphics window. This is an extension of the orbit function ('o' key), but it should orbit the model around a sphere, not just a circle. We will refer to this operation as "trackball rotation". This video shows how this should work.
• When the right mouse button is pressed and the mouse is moved, move the model in the plane of the screen (similar to what the 'x' and 'y' keys did). Scale this operation so that when the model is in the screen space z=0 plane, the model follows your mouse pointer as closely as possible. If you don't have a right mouse button, use your mouse button along with a function key (Shift, Control, Alt, etc.) to enter this mode.
• Use the mouse wheel to move the model along the screen space z axis (i.e., in and out of the screen = push back and pull closer).
• Retain the functionality of the 's'/'S' keys to scale the model about its object space origin. The other keyboard functions for the control of the 3D models are no longer needed, but it does not hurt to keep them supported.

Notes on the Trackball Rotation

The figure below illustrates how to translate mouse movement into a rotation axis and angle. m0 and m1 are consecutive 2D mouse positions. These positions define two locations v and w on an invisible 3D sphere that fills the rendering window. Use their cross product as the rotation axis a = v x w, and the angle between v and w as the rotation angle.

Horizontal mouse movement exactly in the middle of the window should result in a rotation just around the y-axis. Vertical mouse movement exactly in the middle of the window should result in a rotation just around the x-axis. Mouse movements in other areas and directions should result in rotations about an axis a which is not parallel to any single coordinate axis, and is determined by the direction the mouse is moved in.

Once you have calculated the trackball rotation matrix for a mouse drag, you will need to multiply it with the object-to-world transformation matrix of the object you are rotating.

For step by step instructions, take a look at this tutorial. Note that the tutorial was written for Windows messages, instead of GLFW mouse events. This means that you'll need to replace the "CSierpinskiSolidsView::OnLButtonDown", "CSierpinskiSolidsView::OnMouseMove", etc. with an appropriate GLFW equivalent. To help you understand the code better, here is a line-by-line commented version of the trackBallMapping function:

Vec3f CSierpinskiSolidsView::trackBallMapping(CPoint point)    // The CPoint class is a specific Windows class. Either use separate x and y values for the mouse location, or use a Vector3 in which you ignore the z coordinate.
{
    Vec3f v;    // Vector v is the synthesized 3D position of the mouse location on the trackball
    float d;     // this is the depth of the mouse location: the delta between the plane through the center of the trackball and the z position of the mouse
    v.x = (2.0*point.x - windowSize.x) / windowSize.x;   // this calculates the mouse X position in trackball coordinates, which range from -1 to +1
    v.y = (windowSize.y - 2.0*point.y) / windowSize.y;   // this does the equivalent to the above for the mouse Y position
    v.z = 0.0;   // initially the mouse z position is set to zero, but this will change below
    d = v.Length();    // this is the distance from the trackball's origin to the mouse location, without considering depth (=in the plane of the trackball's origin)
    d = (d<1.0) ? d : 1.0;   // this limits d to values of 1.0 or less to avoid square roots of negative values in the following line
    v.z = sqrtf(1.001 - d*d);  // this calculates the Z coordinate of the mouse position on the trackball, based on Pythagoras: v.z*v.z + d*d = 1*1
    v.Normalize(); // Still need to normalize, since we only capped d, not v.
    return v;  // return the mouse location on the surface of the trackball
}

[to be continued]